mlkit-devel — MLKit Developer's Mailing List

[MLKit-devel] mlkit causes Apple's gcc assembler on Snow Leopard to fail

[Tony W. <awu...@ma...>]

I'm trying to make smltojs-4.3.5 on an Apple Macbook running Snow Leopard.
The MLkit make of basis fails.

( osx version 10.6.5, 64-bit extensions YES, gcc version i686-apple-darwin10-gcc-4.2.1, mlton-20100608 compiled from source for x86_64 )
The gcc above is the standard GCC supported by Apple, most recent XcodeTools version 3.2

MLkit has been edited to use OS.Process.isSuccess, so it compiles under MLton.

basis Make fails in "mlkit_basislibs"

from the log:

make mlkit_basislibs
(cd basis && SML_LIB=.. ../bin/mlkit -c -no_gc basis.mlb)
[reading source file:   Initial.sml]
[wrote X86 code file:   MLB/RI/Initial.sml.s]
FATAL:/usr/bin/../libexec/gcc/darwin/x86_64/as: I don't understand '-' flag!
[reading source file:   GENERAL.sml]
[reading source file:   General.sml]
[wrote X86 code file:   MLB/RI/General.sml.s]
FATAL:/usr/bin/../libexec/gcc/darwin/x86_64/as: I don't understand '-' flag!
...

Can one get from MLkit the line which MLkit passes to GCC assembler and repair the code generating this line in MLkit?

Running mlkit with -verbose was tried.
-verbose does not show how the GCC assembler is invoked.

I guess I'm asking for a patch.

Tony

[MLKit-devel] Bug in pattern matching of IntInf values

[Vesa K. <ves...@gm...>]

Consider the following program:

<-- intinf-pat.mlb -->
$(SML_LIB)/basis/basis.mlb
intinf-pat.sml
<-- intinf-pat.mlb -->

<-- intinf-pat.sml -->
fun is0 (0 : IntInf.int) = true
  | is0 _                = false
<-- intinf-pat.sml -->

Compiling it with MLKit (SVN rev 2287) gives the output:

<--->
[reading source file:   intinf-pat.sml]
Impossible: LambdaStatSem.SWITCH_I

(case var1 of 0 => true | _ => false)


LambdaStatEnv: [
  ConEnv: {
   IntInf -> \/().((int list*bool)->intinf),
   :: -> \/('a2).(('a2*'a2 list)->'a2 list),
   nil -> \/('a0).'a0 list,
   false -> \/().bool,
   true -> \/().bool
  };
  TyNameEnv: {
   exn -> [],
   vector -> [],
   ref -> [],
   list -> [nil,::],
   char -> [],
   string -> [],
   real -> [],
   word32 -> [],
   word -> [],
   word8 -> [],
   intinf -> [IntInf],
   int32 -> [],
   int -> [],
   bool -> [true,false]
  };
  LvarEnv: {
   >= -> \/().((intinf*intinf)->bool),
   <= -> \/().((intinf*intinf)->bool),
   > -> \/().((intinf*intinf)->bool),
   < -> \/().((intinf*intinf)->bool),
   abs -> \/().(intinf->intinf),
   mod -> \/().((intinf*intinf)->intinf),
   div -> \/().((intinf*intinf)->intinf),
   - -> \/().(<intinf, intinf>->intinf),
   + -> \/().((intinf*intinf)->intinf),
   * -> \/().(<intinf, intinf>->intinf),
   ~ -> \/().(intinf->intinf),
   toWord31 -> \/().(intinf->word),
   fromWord31 -> \/().(word->intinf),
   toWord8 -> \/().(intinf->word),
   fromWord8X -> \/().(word->intinf),
   fromWord8 -> \/().(word->intinf),
   toWord -> \/().(intinf->word),
   fromWord -> \/().(word->intinf),
   toWord32 -> \/().(intinf->word32),
   fromWordX -> \/().(word->intinf),
   fromWord31X -> \/().(word->intinf),
   fromWord32X -> \/().(word32->intinf),
   fromWord32 -> \/().(word32->intinf),
   fromInt31 -> \/().(int->intinf),
   fromInt32 -> \/().(int32->intinf),
   fromInt -> \/().(int->intinf),
   toInt -> \/().(intinf->int),
   toInt32 -> \/().(intinf->int32),
   toInt31 -> \/().(intinf->int),
   var1 -> \/().intinf,
   is0 -> \/().(intinf->bool)
  };
  ExConEnv: {Bind -> NONE, Match -> NONE, Div -> NONE}
]


let fun is0 var1 = (case var1 of 0 => true | _ => false) in FRAME(is0:
FORALL'a0.'a0) end

Stopping compilation of MLB-file due to error (code 1).
<--->

-Vesa Karvonen

[MLKit-devel] Bug in equality of array types

[Vesa K. <ves...@gm...>]

Consider the following program:

<--- eq.sml --->
fun arrayEq (a : 'a array, b : 'a array) = a = b
fun refEq (a : 'a ref, b : 'a ref) = a = b
<--- eq.sml --->

Compiling it with MLKit (4.3.2) gives:

<--- output --->
[reading source file:   eq.sml]

eq.sml, line 1, column 43:
  fun arrayEq (a : 'a array, b : 'a array) = a = b
                                             ^^^^^
Type clash,
   operand suggests operator type: 'a array * 'a array->'b
   but I found operator type:      ''c * ''c->bool
Stopping compilation of MLB-file due to errors.
<--- output --->

-Vesa Karvonen

Re: [MLKit-devel] Patch to MLKit to compile "Extended Basis Library"

[Martin E. <ma...@it...>]

Hi Vesa,

Thanks for your MLKit improvements. I've applied your patch to the trunk.

Best Regards,

Martin

On 10/28/06, Vesa Karvonen <ves...@cs...> wrote:
> While porting my "Extended Basis Library" project
>
>   http://mlton.org/pipermail/mlton-user/2006-October/000925.html
>
> to MLKit, I made a number of changes to MLKit.  See the patch to
> MLKit here:
>
>   http://mlton.org/cgi-bin/viewsvn.cgi?rev=4787&view=rev
>
> The patch:
> - Exposes the Basis library Text : TEXT module, which wasn't exposed
>   for some reason and also removes the opaque signature ascription which
>   would (incorrectly) generate new types char, string, vector, ...
> - Extends the MLB parser to handle an arbitrary number of annotations
>   in MLB files.  Unrecognized annotations result in a warning and have
>   no other semantic effect.
> - Extends the MLB lexer to recognize simple quoted strings.  This is
>   required to parse MLton's annotations properly.
> - Extends the expansion of MLB path variables so that path variables
>   are allowed at any position of a path rather than just at the
>   beginning.
>
> --Vesa Karvonen
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> MLKit-devel mailing list
> MLK...@li...
> https://lists.sourceforge.net/lists/listinfo/mlkit-devel
>

[MLKit-devel] Patch to MLKit to compile "Extended Basis Library"

[Vesa K. <ves...@cs...>]

While porting my "Extended Basis Library" project

  http://mlton.org/pipermail/mlton-user/2006-October/000925.html

to MLKit, I made a number of changes to MLKit.  See the patch to
MLKit here:

  http://mlton.org/cgi-bin/viewsvn.cgi?rev=4787&view=rev

The patch:
- Exposes the Basis library Text : TEXT module, which wasn't exposed
  for some reason and also removes the opaque signature ascription which
  would (incorrectly) generate new types char, string, vector, ...
- Extends the MLB parser to handle an arbitrary number of annotations
  in MLB files.  Unrecognized annotations result in a warning and have
  no other semantic effect.
- Extends the MLB lexer to recognize simple quoted strings.  This is
  required to parse MLton's annotations properly.
- Extends the expansion of MLB path variables so that path variables
  are allowed at any position of a path rather than just at the
  beginning.

--Vesa Karvonen

[MLKit-devel] Re: Installasjon av MLKit 4.3.0

[Martin E. <ma...@it...>]

Dear Roland,

Sorry for the late reply! For the system-wide installation, simply write

  echo "SML_LIB /usr/lib/mlkit" > /etc/mlkit/mlb-path-map

instead of

  echo "SML_LIB=3D/usr/lib/mlkit" > /etc/mlkit/mlb-path-map   # WRONG

Thanks for your bug-report. The documentation glitch has been fixed in
the source repository.

Best Regards,

Martin Elsman

On 3/3/06, Roland Olsson <ro...@ia...> wrote:
>
>
> Hei,
>
> Jeg har brukt mlton i mange =E5r men tenkte n=E5 pr=F8ve MLKit p=E5 mitt =
prosjekt,
> som inneholder ca 43000 linjer ML kode. Men dessverre oppst=E5r det en
> sannsynligvis triviell feil allerede ved kj=F8ring av "Hello World", neml=
ig
> f=F8lgende.
>
> mlkit helloworld.sml
>
> Error: In file: /home/rolando/.mlkit/mlb-path-map On line 0:Bad sequence
> of tokens in definition of: SML_LIB=3D/home/rolando/mlkit/lib/mlkit.
>
> *** Error: MlbProject.error
>
> Jeg har pr=F8vd b=E5de "system-wide installation" og "personal installati=
on"
> med
>
> echo "SML_LIB=3D/usr/lib/mlkit" > /etc/mlkit/mlb-path-map
>
> respektive
>
> echo "SML_LIB=3D$HOME/mlkit/lib/mlkit" > .mlkit/mlb-path-map
>
>
> Hvordan kan jeg f=E5 MLKit =E5 fungere?
>
>
> Hilsen
>
> Roland
>
>
> --
> Roland Olsson
> F=F8rsteamanuensis, dr.scient - Associate Professor, PhD
> H=F8gskolen i =D8stfold - =D8stfold University College
> Avd. for informasjonsteknologi - Faculty of Computer Sciences
> Tel (+47) 69215369
>
>

[MLKit-devel] svn repository now in place

[Martin E. <ma...@it...>]

Hi,

I have now imported the MLKit sources into a svn (subversion)
repository at sourceforge.net. Source code browsing and information
about source code downloading are available from the MLKit sourceforge
page. For source code downloading, execute the following command from
your shell

  $  svn co https://svn.sourceforge.net/svnroot/mlkit/trunk/kit kit

Comments are welcome.

Best Regards,

Martin

[MLKit-devel] Re: MLKit 4.3.0 Release

[Martin E. <ma...@it...>]

Dear Vesa,

I'm waiting for an SF administrator to copy our cvs-repository from
shell.sourceforge.net into an SF cvs-repository. I have been waiting
since 02-14-06. Hopefully it will be done soon...

Cheers,

Martin

On 2/20/06, Vesa Karvonen <ves...@cs...> wrote:
> Quoting Martin Elsman <ma...@it...>:
> > I've made a fix - and within a week (I hope), you'll be able to see
> > the repository at sourceforge...
> [...]
>
> I noticed that the MLKit project was registed some time ago at SF
> and I registered to the project mailing list.  When/Will the current
> sources appear in the CVS repository of the project?
>
> -Vesa Karvonen
>

[MLKit-devel] does it work

[Martin E. <ma...@it...>]

test